Machine learning 多标签交叉熵_Machine Learning_Neural Network_Multilabel Classification

Machine learning 多标签交叉熵

machine-learning neural-network

Machine learning 多标签交叉熵,machine-learning,neural-network,multilabel-classification,Machine Learning,Neural Network,Multilabel Classification,其中一个答案中存在交叉熵：，即：为什么这是明确的多标签？它看起来很像单变量（单类）分类的对数损失。我在文献中发现了这个 # custom loss: multi label cross entropy def multilabel_objective(predictions, targets): epsilon = np.float32(1.0e-6) one = np.float32(1.0) pred = T.clip(predictions, epsilon, o

其中一个答案中存在交叉熵：，即：

为什么这是明确的多标签？它看起来很像单变量（单类）分类的对数损失。我在文献中发现了这个

# custom loss: multi label cross entropy
def multilabel_objective(predictions, targets):
    epsilon = np.float32(1.0e-6)
    one = np.float32(1.0)
    pred = T.clip(predictions, epsilon, one - epsilon)
    return -T.sum(targets * T.log(pred) + (one - targets) * T.log(one - pred), axis=1)